Individual Health Insurance, Bluhm, William and Leida, Hans, 2nd Edition - Ch. 4: Managing Antiselection (pp. 109-148)

Use this thread to discuss ANYTHING and EVERYTHING related to this syllabus reading.
Some possible questions include:

  • How can this reading be tested?
  • I don’t understand a specific topic/formula - Can we discuss this?
  • This reading gives me nightmares. Can we talk through it a bit?

Good luck!

Anyone want to discuss the premium leakage and buy-down effect calculations? The example in the manual (I’m using MATE) seems pretty straight foward, however this was tested on exam GHA spring 2018 question 8, subquestion c. In the model solution, I don’t understand why they didn’t apply the 20% rate increase when calculating the buy-down effect.

In the manual, the “Actual premium before buy down” takes into consideration the rate increase. Do you guys think the approach in MATE is the right one, and that the model solution is incorrect, or am I missing something here?

I find buy down and premium leakage concepts very difficult to explain concisely, but I’ll do my best.

First of all, I think the model solution does actually take into consideration the rate increase, although I would agree that the way the solution is presented is confusing.

For me it feels more natural to calculate the buy down effect first. If the solution would have shown their work better, it would have been more obvious that the $180 comes from the original $150 premium with a 20% increase. The $171.18 is the expected premium with shift assuming the leaner benefit costs 7% less (i.e. 70% shift * 93% * $180 + 30% persist * $180). This is lower than $180 because members are obviously buying down their benefits as given in the problem. The $8.82 difference is the buy down effect, which I think of as the “expected premium without any shift” minus the “expected premium with shift.” The problem in practice is that the “expected premium with shift” doesn’t take into account the difference in claims due to risk – that part is rather unexpected.

I think of premium leakage as the “actual premium with shift” minus the “expected premium with shift” (same as above $171.18). The “actual premium with shift” is not the premium that will be received, but rather the premium desired by taking into account the actual difference in risk profile between shifting & persisting members. This is calculated using the expected claims, i.e. 70% shift * $150 * 93% + 30% persist * $250 = $172.65. There is no 20% increase applied to claims here, because the problem did not say claims were expected to increase – only premiums. The difference is then the premium leakage, $172.65 - $171.18 = $1.47.

Hope that helps. Feel free to question or correct anything I may be misunderstanding.


Wow thanks so much for your detailed response!

I got confused with the $180, didn’t realize that it was from the $150 with the 20% premium increase, I was looking at the expected claims (which don’t need to be increased by 20%).

It makes much more sense now, thanks again. I think for me to really remember it better, I would flip it around and say the buy-down effect is actual premium without shit - actual premium with shift (so you calculate it with the actual premium that is paid, not the claims) and then the premium leakage would be the expected premium with shift (sort of like an expected value that you get from claims, and add expenses that are 0 in that problem) and reduce that by actual premium with shift. So yeah your methodology of calculating the buy-down effect first makes a lot of sense!
I’ll just make sure I use the terms that help me calculate the right value for each one!

Thanks again for you help!

Are ACA’s three R’s premium stabilization programs on the syllabus? Info is somewhat conflicting regarding this when looking at the MATE manual and the syllabus description.

That’s a very good explanation! Key insight is that the first term in the “premium leakage” equation has to do with the claims desired. For these problems I just write out the equations and solve for the terms (as long as you understand exactly what the those terms are). There are only 3 terms to solve for…